Emotion detection has a bright future in store for marketers

“If we don’t make you cry, we fail.” Lorraine Twohill, Google's top marketing boss, has no doubt about it: marketing is about emotion. Wouldn’t it be nice if our marketing systems would recognise feelings automatically?

Marketers have become very familiar with the term contextual marketing. It roughly means that you adjust the marketing message to the current circumstances (the context): the location, user preferences, the device. Emotion is an essential ingredient of this mix. That is not as crazy as it sounds. You don’t approach a gloomy person in the same way after all as you would someone who is ecstatic.

Emotion plays a very important role in marketing. Research shows that emotion driven marketing works twice as well as more traditional marketing. Purchase decisions are primarily driven by emotions, it is only after the fact that we try to rationalise a purchase. Without emotion it is very difficult to tie people to your brand for a longer period of time. Technology that recognises emotions, may adjust the marketing message accordingly and increase the chance that the interested party proceeds with the purchase.

Marketers who unleash this technology completely automatically on their ‘target group' have much higher chances of success than their colleagues who don't. No wonder that the market of automated emotion recognition is estimated to be worth around 23 billion dollars in 2020.

Apple

Apple is a good example of a company that recognises the value of emotion in the marketing mix. In 2014 they already submitted a patent that describes how software can analyse the emotion of users based on facial expression.

At the start of 2016 Apple took over Emotient, a start-up specialised in emotion recognition with the aid of artificial intelligence18> Their machine learning algorithm analyses up to 100,000 facial expressions per day, enabling the artificial intelligence to recognise moods increasingly more accurately. It is not inconceivable that this technology will show up on iPads and iPhones to provide all new kinds of interactions. Display advertisements could for example be adjusted to your mood, or Siri could talk to you in a slightly different way depending on whether you are cranky or cheerful.

Chatbots are at any rate an important application area of emotion detection for marketers. Based on machine learning trained algorithms they know how to approach angry, disappointed or unhappy customers appropriately. This is of course without customers being aware that they are talking to a robot and they won’t get the feeling that they’re not being taken seriously.

Facebook

Display advertising is another important application of emotion detection. Facebook has been investing heavily in emotion detection. One year ago the like button was replaced by a variant with five emoticons to indicate what sentiments you have with a message. This is valuable data because Facebook will adjust the newsfeed on your timeline on this basis. The system is currently still fairly primitive but this will undoubtedly change. It is my prediction that you won’t only be able to segment Facebook ads on age, location and job, but also on emotion.

Monitoring

Emotion recognition is also valuable for brand and social media monitoring. A lot of tools for social media monitoring can already recognise emotions regarding your brand, by means of text recognitions and Language Processing (NLP).

Most tools are currently using a list of ‘alert words' that divide reactions and posts into positive, neutral or negative. Algorithms for text recognition are nevertheless improving quickly. When a system recognises a text better, it can estimate the underlying sentiment much more accurately.

Viral or not?

Emotion detection is not only useful for marketers for automated real-time interaction with the target group. It also demonstrates its value for marketing campaigns. Multiple brands are working with analysis software that measures emotions in test audiences. It enables them to determine beforehand which campaigns will stir up the most desirable emotions, so it has a better chance of going viral.

Kellogg’s for example used emotion detection software to choose the most effective YouTube advertisement. The manufacturer showed the trial audience a number of versions for new breakfast cereal. Animals and other creatures played a large part in it. The version with a snake created the most laughs, but little commitment. With the aid of facial recognition technology they ended up choosing for a version with an alien. This ad brought about positive emotions and a lot of interaction.

https://www.youtube.com/watch?v=4hiYrOiGaRg

Value of wearables

Emotions are not only discerned from language usage, facial expressions or clicking on certain emoticons. Several biometric signals such as heart rate, skin hydration and blood pressure may give away emotions such as stress, happiness and grief. Wearables are in this regard the ideal ‘feelers’ for emotion detection systems.

The first examples of this are showing up now. The start-up Sentio Solutions recently presented ’Feel’ a wrist band that recognises your emotions on the basis of your ‘Galvanic Skin Response’. A sensor measures how your skin reacts to small electric pulses, which in turn provides information about hydration. The wrist band will combine that data with blood pressure, movement, and heart rate and tries to determine your mood.

The applications are literally endless: the creators of Feel want to release an API for developers so that everyone can write software that uses the measured emotions. They themselves give a link to Spotify as an example: the music service could create an API that recommends songs with certain moods based on your music taste. We’re waiting for the first application for marketers.

Not just marketing

Automated emotion recognition is not just interesting for the marketing field. The Transport Security Administration (TSA), the organisation ensuring security in American airports, uses emotion recognition as a security strategy. According to their SPOT method security officers scan people for nervous ticks and behavior, which may signal bad intentions. The security officers performing this task are for now still human, but the step to an automated algorithm is small due to the methodological approach.

Systems for emotion recognition already exist, and we will only see further developments in the future. They allow for a strongly personalised customer approach, customer experience and optimisation of marketing campaigns. This can only be good news for marketers.